feat(scraper): add RWEB best-practices waves 1 and 2

components/ecoindex/best_practices/default_rules.yaml

@@ -62,3 +62,134 @@ rules:
     higher_is_worse: false
     thresholds:
       fail: 1
+
+  - id: stylesheets_number
+    enabled: true
+    checker: stylesheets_number
+    category: network
+    rweb_id: RWEB_0035
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0035-limit-the-number-of-css
+    title: "Limit the number of CSS"
+    description: >
+      Limit stylesheet files to avoid multiplying HTTP requests and simplify
+      browser rendering. Prefer a shared CSS plus a few media-specific sheets.
+    thresholds:
+      # GreenIT-Analysis: >7 warn, >10 fail
+      warn: 7
+      fail: 10
+
+  - id: standard_typefaces
+    enabled: true
+    checker: standard_typefaces
+    category: network
+    rweb_id: RWEB_0032
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0032-prefer-standard-fonts
+    title: "Prefer standard fonts"
+    description: >
+      Prefer system fonts already available on the user device. Downloaded
+      web fonts consume bandwidth and delay text rendering.
+    thresholds:
+      warn: 1
+      fail: 2
+
+  - id: social_network_button
+    enabled: true
+    checker: social_network_button
+    category: user-device
+    rweb_id: RWEB_0059
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0059-replace-the-official-social-network-sharing-buttons
+    title: "Replace official social network sharing buttons"
+    description: >
+      Official social widgets load heavy third-party scripts. Prefer lightweight
+      share links that do not pull external SDKs.
+    thresholds:
+      fail: 0
+
+  - id: animated_gifs
+    enabled: true
+    checker: animated_gifs
+    category: user-device
+    rweb_id: RWEB_0099
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0099-limit-the-use-of-animated-gif
+    title: "Limit the use of animated GIFs"
+    description: >
+      Animated GIFs are heavier and less efficient than modern video formats
+      such as WebM or MP4 for animations.
+    thresholds:
+      fail: 0
+
+  - id: http_errors
+    enabled: true
+    checker: http_errors
+    category: network
+    title: "Avoid HTTP error responses"
+    description: >
+      Resources returning HTTP status codes greater than or equal to 400 waste
+      bandwidth and often indicate broken dependencies.
+    thresholds:
+      fail: 0
+
+  - id: http_redirects
+    enabled: true
+    checker: http_redirects
+    category: network
+    rweb_id: RWEB_0112
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0112-avoir-redirections
+    title: "Avoid redirects"
+    description: >
+      Redirects degrade response time and consume resources. Prefer direct URLs
+      whenever possible.
+    thresholds:
+      fail: 1
+
+  - id: http2
+    enabled: true
+    checker: http2
+    category: network
+    rweb_id: RWEB_0083
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0083-http2-over-http1
+    title: "Prefer HTTP/2 to HTTP/1"
+    description: >
+      HTTP/2 multiplexes resources over a single connection and transfers fewer
+      bytes than HTTP/1, improving load time and efficiency.
+    thresholds:
+      fail: 0
+
+  - id: cache_headers
+    enabled: true
+    checker: cache_headers
+    category: network
+    rweb_id: RWEB_0075
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0075-add-expires-or-cache-control-headers
+    title: "Add Expires or Cache-Control headers"
+    description: >
+      Static resources should declare cache lifetime via Cache-Control or
+      Expires so browsers avoid re-downloading unchanged assets.
+    thresholds:
+      fail: 0
+
+  - id: http_compression
+    enabled: true
+    checker: http_compression
+    category: network
+    rweb_id: RWEB_0076
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0076-compressing-text-files-css-js-html-and-svg
+    title: "Compress CSS, JavaScript, HTML and SVG files"
+    description: >
+      Text resources should be served with HTTP compression (gzip, brotli, …)
+      to reduce transferred bytes.
+    thresholds:
+      fail: 0
+
+  - id: no_cookie_static
+    enabled: true
+    checker: no_cookie_static
+    category: datacenter
+    rweb_id: RWEB_0081
+    url: https://rweb.greenit.fr/en/fiches/RWEB_0081-host-static-resources-on-a-cookie-free-domain
+    title: "Host static resources on a cookie-free domain"
+    description: >
+      Cookies attached to static asset requests inflate every transfer. Serve
+      CSS, JS, images and fonts from a cookie-less domain.
+    thresholds:
+      fail: 1

components/ecoindex/best_practices/har_utils.py

@@ -0,0 +1,159 @@
+"""Helpers inspired by GreenIT-Analysis HAR resource classification."""
+
+from __future__ import annotations
+
+import re
+from datetime import datetime
+from urllib.parse import urlparse
+
+from ecoindex.models.scraper import RequestItem
+
+HTTP_REDIRECT_CODES = frozenset({301, 302, 303, 307})
+HTTP_COMPRESSION_TOKENS = frozenset(
+    {"br", "compress", "deflate", "gzip", "pack200-gzip"}
+)
+
+_STATIC_CATEGORIES = frozenset(
+    {"image", "javascript", "font", "css", "audio", "video"}
+)
+_COMPRESSIBLE_CATEGORIES = frozenset({"css", "javascript", "html", "font"})
+
+_FONT_MIME_RE = re.compile(
+    r"(font/|application/(?:x-)?font-|application/vnd\.ms-fontobject|"
+    r"application/font-sfnt)",
+    re.I,
+)
+_COMPRESSIBLE_MIME_RE = re.compile(
+    r"^(text/|application/(?:javascript|x-javascript|json|xml|xhtml\+xml|"
+    r"ld\+json|rss\+xml|atom\+xml|manifest\+json)|image/svg\+xml|"
+    r"font/(?:eot|opentype))",
+    re.I,
+)
+_STATIC_MIME_RE = re.compile(
+    r"^(image/|audio/|video/|text/css|text/javascript|application/javascript|"
+    r"application/x-javascript|font/|application/(?:x-)?font-|"
+    r"application/vnd\.ms-fontobject|application/pdf|application/octet-stream|"
+    r"application/zip|application/x-shockwave-flash|application/manifest\+json)",
+    re.I,
+)
+
+_SOCIAL_PATTERNS: list[tuple[str, str]] = [
+    ("platform.twitter.com/widgets.js", "twitter"),
+    ("platform.linkedin.com/in.js", "linkedin"),
+    ("assets.pinterest.com/js/pinit.js", "pinterest"),
+    ("platform-api.sharethis.com/js/sharethis.js", "sharethis"),
+    ("s7.addthis.com/js/300/addthis_widget.js", "addthis"),
+    ("static.addtoany.com/menu/page.js", "addtoany"),
+]
+
+
+def header_map(headers: list[dict] | None) -> dict[str, str]:
+    result: dict[str, str] = {}
+    if not headers:
+        return result
+    for header in headers:
+        name = header.get("name")
+        value = header.get("value")
+        if name is None or value is None:
+            continue
+        result[str(name).lower()] = str(value)
+    return result
+
+
+def is_font_resource(item: RequestItem) -> bool:
+    if item.category == "font" or _FONT_MIME_RE.search(item.mime_type or ""):
+        return item.size > 0
+    mime = (item.mime_type or "").split(";")[0].strip().lower()
+    if mime in {"", "text/plain", "application/octet-stream"}:
+        path = urlparse(item.url).path.lower()
+        if path.endswith((".woff", ".woff2", ".ttf", ".otf", ".eot")):
+            return item.size > 0
+        if ".woff?" in item.url.lower() or ".woff2?" in item.url.lower():
+            return item.size > 0
+    return False
+
+
+def is_css_resource(item: RequestItem) -> bool:
+    return item.category == "css" or "css" in (item.mime_type or "").lower()
+
+
+def is_gif_resource(item: RequestItem) -> bool:
+    mime = (item.mime_type or "").lower()
+    if "gif" in mime:
+        return True
+    path = urlparse(item.url).path.lower()
+    return path.endswith(".gif")
+
+
+def is_static_resource(item: RequestItem) -> bool:
+    if item.category in _STATIC_CATEGORIES:
+        return True
+    return bool(_STATIC_MIME_RE.search((item.mime_type or "").split(";")[0].strip()))
+
+
+def is_compressible_resource(item: RequestItem) -> bool:
+    if item.size <= 150:
+        return False
+    if item.category in _COMPRESSIBLE_CATEGORIES:
+        return True
+    return bool(
+        _COMPRESSIBLE_MIME_RE.search((item.mime_type or "").split(";")[0].strip())
+    )
+
+
+def is_resource_compressed(item: RequestItem) -> bool:
+    encoding = (item.content_encoding or "").lower().strip()
+    if not encoding:
+        return False
+    # Content-Encoding can be a comma-separated list.
+    tokens = {token.strip() for token in encoding.split(",")}
+    return bool(tokens & HTTP_COMPRESSION_TOKENS)
+
+
+def has_valid_cache_headers(item: RequestItem) -> bool:
+    cache_control = item.cache_control or ""
+    is_valid = False
+    if cache_control and not re.search(
+        r"(no-cache)|(no-store)|(max-age\s*=\s*0)", cache_control, re.I
+    ):
+        is_valid = True
+
+    if item.expires:
+        expires = _parse_http_date(item.expires)
+        if expires is not None:
+            now = _parse_http_date(item.response_date) or datetime.now(tz=expires.tzinfo)
+            if expires < now:
+                is_valid = False
+
+    return is_valid
+
+
+def _parse_http_date(value: str | None) -> datetime | None:
+    if not value:
+        return None
+    try:
+        from email.utils import parsedate_to_datetime
+
+        return parsedate_to_datetime(value)
+    except (TypeError, ValueError, IndexError):
+        return None
+
+
+def is_http1(item: RequestItem) -> bool:
+    version = (item.http_version or "").upper()
+    return "HTTP/1" in version
+
+
+def is_http_redirect(status: int) -> bool:
+    return status in HTTP_REDIRECT_CODES
+
+
+def official_social_network(url: str) -> str | None:
+    lower = url.lower()
+    host = (urlparse(url).hostname or "").lower()
+    if host == "connect.facebook.net" and "sdk.js" in lower:
+        return "facebook"
+    for pattern, name in _SOCIAL_PATTERNS:
+        if pattern in lower:
+            return name
+    return None

components/ecoindex/best_practices/models.py

@@ -14,6 +14,7 @@ class RuleCategory(str, Enum):
 
     network = "network"
     user_device = "user-device"
+    datacenter = "datacenter"
 
 
 class Thresholds(BaseModel):